test-nginx.py: added new testHEAD master

author: Marc Deslauriers <marc.deslauriers@canonical.com> 2026-07-29 11:10:50 -0400
committer: Marc Deslauriers <marc.deslauriers@canonical.com> 2026-07-29 11:10:50 -0400
commit: ffd9fdd1d2a52f0f072fb6f92b7775443e44fdb2 (patch)
tree: a53e33cf93d39de587a1f5d0319461dffa30312b /scripts
parent: ccd39b42014d3126113569824ab25fa8e62b8e41 (diff)
1 files changed, 211 insertions, 0 deletions
diff --git a/scripts/test-nginx.py b/scripts/test-nginx.py
index 05ff3389..7d58739e 100755
--- a/scripts/test-nginx.py
+++ b/scripts/test-nginx.py
@@ -67,6 +67,11 @@ class NginxTest(testlib_httpd.HttpdCommon):
         self.ssl_crt = "/etc/nginx/cert.pem"
         self.ssl_key = "/etc/nginx/cert.key"
 
+        # conf.d files are included in the http context by the default
+        # nginx.conf, so they are suitable for directives such as `map`
+        # that are only valid in the http context.
+        self.conf_d = "/etc/nginx/conf.d"
+
         self.pidfile = "/var/run/nginx.pid"
 
         self.daemon = testlib.TestDaemon("nginx")
@@ -81,6 +86,12 @@ class NginxTest(testlib_httpd.HttpdCommon):
         if os.path.exists(self.ssl_key):
             os.unlink(self.ssl_key)
 
+        # Remove any conf.d snippets created by individual tests so they
+        # don't leak into subsequent tests or the running configuration.
+        for filename in os.listdir(self.conf_d):
+            if filename.startswith("qrt-") and filename.endswith(".conf"):
+                os.unlink(os.path.join(self.conf_d, filename))
+
         testlib.config_restore(self.default_vhost)
 
     def _test_headers(self, url="https://localhost/", content="", invert=False, check_ssl=True):
@@ -258,6 +269,206 @@ server {
         self._test_headers(url="http://localhost/", content="X-QRT-Test: headers-more-works", check_ssl=False)
         self._test_headers(url="http://localhost/", content="Server:", invert=True, check_ssl=False)
 
+    def _get_nginx_worker_pids(self):
+        '''Return the set of current nginx worker process PIDs.'''
+        rc, report = testlib.cmd(['pgrep', '-x', 'nginx'])
+        # pgrep matches both master and workers; exclude the master pid.
+        master_pid = None
+        try:
+            with open(self.pidfile, 'r') as fd:
+                master_pid = fd.readline().rstrip('\n')
+        except (IOError, OSError):
+            pass
+        pids = set()
+        if rc == 0:
+            for line in report.strip().split('\n'):
+                pid = line.strip()
+                if pid and pid != master_pid:
+                    pids.add(pid)
+        return pids
+
+    def test_cve_2026_42533(self):
+        '''Verify CVE-2026-42533 (heap buffer overflow with map regex)'''
+
+        # CVE-2026-42533: a heap buffer overflow could occur in a worker
+        # process when using the `map` directive with regex matching if the
+        # map variable was included in a string expression after a regex
+        # capture affected by that same map. A similar issue could happen
+        # when using a non-cacheable (`volatile`) variable in a string
+        # expression.
+        #
+        # The nginx script engine evaluates complex values (string
+        # expressions with variables) in two passes: a length pass to
+        # calculate the buffer size, then a copy pass to fill it. When a
+        # map variable with regex matching is evaluated, it runs the regex
+        # and sets r->captures as a side effect. If a regex capture ($1,
+        # $named_capture) appears in the same expression BEFORE the map
+        # variable, the length pass uses the original captures, but by the
+        # copy pass the captures have been overwritten by the map's regex —
+        # causing the copy to write more data than the buffer can hold.
+        #
+        # The fix (nginx 1.31.3, 1.30.4) adds buffer boundary checking in
+        # the script engine (ngx_http_script_check_length) that logs "no
+        # buffer space in script copy" and returns a 500 error instead of
+        # overflowing.
+        #
+        # On a vulnerable build the worker process crashes (SIGSEGV); on a
+        # fixed build the worker survives and returns an error response.
+        #
+        # This test has two parts:
+        #   1. Trigger test: exercise the exact CVE-2026-42533 code path and
+        #      verify the worker does not crash.
+        #   2. Functionality test: verify normal map regex and volatile map
+        #      usage produces correct values (no regression from the fix).
+
+        # The `map` directive is only valid in the http context, so place it
+        # in a conf.d snippet (included by the default nginx.conf) rather
+        # than in the server block.
+        map_conf = os.path.join(self.conf_d, "qrt-cve-2026-42533.conf")
+        testlib.create_fill(map_conf, '''
+# CVE-2026-42533 regression test
+#
+# Trigger 1: map with regex matching where the map value uses a named
+# capture from that same regex. When the map variable is used in a string
+# expression after a capture reference, the map's regex execution
+# overwrites r->captures between the length and copy passes.
+map $uri $qrt_map {
+    default                "unknown";
+    "~^(?P<qrt_capture>.+)$"  "$qrt_capture";
+}
+
+# Trigger 2: non-cacheable (volatile) map. A volatile variable is
+# re-evaluated on every access, so using it in a string expression can
+# cause the length to change between the two passes.
+map $request_uri $qrt_volatile {
+    volatile;
+    default                "default-volatile";
+    "~^/volatile"          "volatile-matched";
+}
+''')
+
+        # --- Part 1: Trigger test (worker crash detection) ---
+        #
+        # The trigger uses the rewrite module's `set` directive to create a
+        # string expression where a capture appears before the map variable.
+        # `set $qrt_temp "$qrt_capture $qrt_map"` evaluates $qrt_capture
+        # (empty, since no location regex set it) then $qrt_map (which runs
+        # the map regex, overwriting r->captures). On the copy pass,
+        # $qrt_capture reads from the now-modified captures — writing far
+        # more bytes than the 0 allocated for it → heap buffer overflow.
+        #
+        # A long URI is used so the overflow is large enough to crash the
+        # worker process on a vulnerable build.
+        testlib.config_replace(self.default_vhost, '''
+server {
+	listen 80;
+	server_name localhost;
+
+	root %s;
+	index index.html index.htm index.nginx-debian.html;
+
+	location / {
+		set $qrt_capture "";
+		set $qrt_temp "$qrt_capture $qrt_map";
+		return 200 "$qrt_temp\\n";
+	}
+
+	location = /volatile {
+		return 200 "volatile-endpoint-ok\\n";
+	}
+}
+''' % self.document_root, append=False)
+
+        rc, report = self.daemon.restart()
+        self.assertTrue(rc, report)
+        self.assertTrue(testlib.check_pidfile("nginx", self.pidfile))
+
+        # Record worker PIDs before sending the trigger request.
+        pids_before = self._get_nginx_worker_pids()
+        self.assertTrue(len(pids_before) > 0,
+                        "No nginx worker processes found")
+
+        # Send a request with a long URI to trigger the overflow. On a
+        # vulnerable build, the worker crashes (segfault) and the master
+        # restarts it with a new PID. On a fixed build, the buffer overrun
+        # protection catches the overflow and returns a 500 error, but the
+        # worker stays alive.
+        long_uri = "/" + "A" * 4096
+        testlib.cmd(['curl', '-s', '-o', '/dev/null',
+                     'http://localhost' + long_uri])
+
+        # Worker PIDs should be unchanged — the same workers are still
+        # running. On a vulnerable build, the worker would have crashed and
+        # been restarted with a different PID.
+        pids_after = self._get_nginx_worker_pids()
+        self.assertEqual(pids_before, pids_after,
+                         "nginx worker process crashed and was restarted "
+                         "(PIDs changed: %s -> %s). This indicates "
+                         "CVE-2026-42533 is not fixed." %
+                         (pids_before, pids_after))
+
+        # --- Part 2: Functionality test (no regression) ---
+        #
+        # Replace the vhost config with normal map usage (no trigger) to
+        # verify the fix didn't break legitimate map regex and volatile
+        # functionality.
+        testlib.config_replace(self.default_vhost, '''
+server {
+	listen 80;
+	server_name localhost;
+
+	root %s;
+	index index.html index.htm index.nginx-debian.html;
+
+	# Map output used in a string expression (normal functionality).
+	# Use `return` instead of `add_header` because add_header is not
+	# applied to 404 responses, and we want to verify the map value
+	# directly regardless of whether a file exists on disk.
+	location / {
+		return 200 "mapped:$qrt_map vol:$qrt_volatile\\n";
+	}
+}
+''' % self.document_root, append=False)
+
+        rc, report = self.daemon.restart()
+        self.assertTrue(rc, report)
+        self.assertTrue(testlib.check_pidfile("nginx", self.pidfile))
+
+        # Regex map: a request URI should be captured by the map regex.
+        # The volatile map matches /testpath against default (no /volatile
+        # prefix), so we expect "default-volatile".
+        rc, report = testlib.cmd(['curl', '-s',
+                                  'http://localhost/testpath'])
+        expected = 0
+        result = 'Got exit code %d, expected %d\n' % (rc, expected)
+        self.assertEqual(expected, rc, result + report)
+        self._word_find(report, "mapped:/testpath")
+        self._word_find(report, "vol:default-volatile")
+        # Worker must still be alive.
+        self.assertTrue(testlib.check_pidfile("nginx", self.pidfile))
+
+        # Regex map: a different path should produce a different capture.
+        rc, report = testlib.cmd(['curl', '-s',
+                                  'http://localhost/another'])
+        expected = 0
+        result = 'Got exit code %d, expected %d\n' % (rc, expected)
+        self.assertEqual(expected, rc, result + report)
+        self._word_find(report, "mapped:/another")
+        # Worker must still be alive.
+        self.assertTrue(testlib.check_pidfile("nginx", self.pidfile))
+
+        # Volatile map: matching request URI should produce the matched
+        # value, and the volatile (non-cacheable) code path should work.
+        rc, report = testlib.cmd(['curl', '-s',
+                                  'http://localhost/volatile'])
+        expected = 0
+        result = 'Got exit code %d, expected %d\n' % (rc, expected)
+        self.assertEqual(expected, rc, result + report)
+        self._word_find(report, "mapped:/volatile")
+        self._word_find(report, "vol:volatile-matched")
+        # Worker must still be alive.
+        self.assertTrue(testlib.check_pidfile("nginx", self.pidfile))
+
     def test_proxy(self):
         '''Test proxy'''
author	Marc Deslauriers <marc.deslauriers@canonical.com>	2026-07-29 11:10:50 -0400
committer	Marc Deslauriers <marc.deslauriers@canonical.com>	2026-07-29 11:10:50 -0400
commit	ffd9fdd1d2a52f0f072fb6f92b7775443e44fdb2 (patch)
tree	a53e33cf93d39de587a1f5d0319461dffa30312b /scripts
parent	ccd39b42014d3126113569824ab25fa8e62b8e41 (diff)